[ttan98]

Rick,

I am a casual visitor to your audio circle and I notice you have designed a number of beautifully made speakers and I am sure all of them sound very good.

I am a novice speaker designer and I notice the FR of most of your speakers are not really ruler flat. I further notice that there is dip around 2.5kHz and starts rising again to reach the same level around 3.5kHz and rise 1-2dB above the midband frequencies. This phenomena occurs in the Verita model and occur in somewhat similar pattern in one or 2 other speakers designed by you, eg S6. Furthermore I also notice the Dynaudio Excite X12(very well received by a few reviewers) has a similar dip around 2kHz.

My question is this a mere coincident in design or on purpose ie you design a 2kHz dip because human is sensitive around that region. If it is on purpose please explain why you do that. This is my observation and nothing more than just that.

Thanks for your time.

[Rick Craig]

Rick,

I am a casual visitor to your audio circle and I notice you have designed a number of beautifully made speakers and I am sure all of them sound very good.

I am a novice speaker designer and I notice the FR of most of your speakers are not really ruler flat. I further notice that there is dip around 2.5kHz and starts rising again to reach the same level around 3.5kHz and rise 1-2dB above the midband frequencies. This phenomena occurs in the Verita model and occur in somewhat similar pattern in one or 2 other speakers designed by you, eg S6. Furthermore I also notice the Dynaudio Excite X12(very well received by a few reviewers) has a similar dip around 2kHz.

My question is this a mere coincident in design or on purpose ie you design a 2kHz dip because human is sensitive around that region. If it is on purpose please explain why you do that. This is my observation and nothing more than just that.

Thanks for your time.

Actually my designs fit a fairly tight frequency window but the on-axis curve really doesn't tell you what the speaker will sound like. I'm more interested in how linear it is over several angles, especially in the horizontal plane. I overlap the on-axis and off-axis curves (30 degrees) to show what the total coverage is. I also take 1/12 octave measurements for floorstanders/monitors which are much better than the common 1/3 octave graphs you see everywhere.

If you take a look at some of the NRC measurements on Soundstage you'll see that many speakers with a flat on-axis response in the 2-3K range tend to have a peak off-axis horizontally in the same region. Unfortunately that's where your ear is most sensitive. It also can be an area where sub par recordings have problems.

The voicing for power response also comes into play and that's really a matter of taste. You can have two designs that have a tight response window of the same, say +/-2db, but one is tilted upward and the other downward.

[ttan98]

Thanks for the reply,

The image is from Dynaudio excite 12, the freq. variation is slight over 15 to 30 deg off axis, this I assume is what you mean, furthermore, the dip at 2khz is a matter of taste and does not matter that much.

Another issue I found that most well received/reviewed 2way bookshelf speakers have a peak at 100Hz and a valley between 100Hz to 1kHz like the image attached.

So if we design a 2-way speaker in accordance to these 2 criteria(and a few others like smooth impedance and minimum phase variations(impt?)) then we have a pair of good sounding speakers.

Also refer to the image, the peak at 750Hz can we hear it?

[Rick Craig]

Thanks for the reply,

The image is from Dynaudio excite 12, the freq. variation is slight over 15 to 30 deg off axis, this I assume is what you mean, furthermore, the dip at 2khz is a matter of taste and does not matter that much.

Another issue I found that most well received/reviewed 2way bookshelf speakers have a peak at 100Hz and a valley between 100Hz to 1kHz like the image attached.

So if we design a 2-way speaker in accordance to these 2 criteria(and a few others like smooth impedance and minimum phase variations(impt?)) then we have a pair of good sounding speakers.

Also refer to the image, the peak at 750Hz can we hear it?

Well the Dynaudio differs in that the 2K dip is at all angles whereas our off-axis curves are flatter. The 750hz area looks more troublesome though and that would definitely create some coloration. I would check out some of the work Floyd Toole and Sean Olive have done at the NRC and Harman.

[Napalm]

The dip at 2kHz shows poor integration at crossover frequency.

The boost in the 100-200Hz range is usually done intentionally, by design, for small speakers that are intended for (stereo) music reproduction without a subwoofer. It hides the lack of acoustic energy in the 0-80Hz range and balances the overall tonality. If you target your speakers for integration with subwoofer, don't do it (aim for ruler flat response).

nap. 

[Rick Craig]

The dip at 2kHz shows poor integration at crossover frequency.

The boost in the 100-200Hz range is usually done intentionally, by design, for small speakers that are intended for (stereo) music reproduction without a subwoofer. It hides the lack of acoustic energy in the 0-80Hz range and balances the overall tonality. If you target your speakers for integration with subwoofer, don't do it (aim for ruler flat response).

nap. 

Not necessarily poor integration - it could also be cabinet diffraction and/or a dip in the driver's response. The 100hz peak can be intentional as you noted given the poor extension of this speaker; however, it can also be a result of the crossover.

[Napalm]

I just took a look at the Verita FR. Impressive. I guess the crossover is at 2.5-2.7 kHz?

As a personal preference I like charts that go the whole spectrum (not divided in bass / mids+treble).

Nap. 

[Rick Craig]

I just took a look at the Verita FR. Impressive. I guess the crossover is at 2.5-2.7 kHz?

As a personal preference I like charts that go the whole spectrum (not divided in bass / mids+treble).

Nap. 

Sorry but the crossover point is proprietary. 

The MLS gate works best for split curves (200hz-20K MLS and 10-200hz nearfield sine) when measuring indoors as I do. The full spectrum graphs are better in a large anechoic chamber or outdoors where you can get better low frequency resolution.

[Napalm]

Sorry but the crossover point is proprietary. 

No problem, I don't build speakers so I can live without knowing 

Nap.